3. Coastal region. 3.1 Geology and palaeogeography of the coastal region

Transcription

1 3. Coastal region 3.1 Geology and palaeogeography of the coastal region Introduction This paragraph, primarily based on Mol (2006), summarises the geology and palaeogeography of the sites Ypenburg, Wateringen 4, Schipluiden, Sion and Rijswijk-A4. Together these sites span an area at the inland side of a seaward prograding beach barrier system, occupied between c and 3380 BC. Pleistocene deposits occur at relative large depths in this area and had been buried by metres of sediments in the first part of the Holocene (until c BC). The Holocene subsurface of the region consists of fluvial deposits, tidal-flat deposits, tidal channel deposits, lagoon deposits and estuarine deposits consisting of sand, clay and peat. The coast was characterised by erosion during the first part of the Holocene. As a result, Mesolithic sites west of the study area may have been eroded. The oldest preserved beach deposits are those occupying a most easterly position and are dated to c BC (Cleveringa 2000). Their preservation marks a switch from transgressive coastal formation into progradational beach barrier formation. This switch is related to the decreased rise of the sea level (globally) on the one hand, and plenty of available sediment at a shallow depth in the coastal zone on the other hand (Beets and Van der Spek 2000). During the Middle Holocene, a beach barrier system formed by c BC protected the region from direct marine influence (Mol and Louwe Kooijmans in prep.; Van der Valk 2006). Isolated low coastal dunes developed on top of the beach barriers (c BC), which gradually healed to form larger barrier complexes. Behind the beach barriers, former tidal flats and tidal channels transformed into sandy beach plains. During storms, the sea incidentally flooded these plains, resulting in wash-over deposits. Low dunes developed on top of the wash-over deposits and beach plains. While relative sea-level rise continued at slowed-down rates, westward progradation, healing and closure of the barrier system transformed the area into a protected lagoon. In that situation, freshwater peat started to form in the study area at c BC, indicative of the gradual rise of the local ground water table. Figure 3.1 shows a palaeogeographical map from the region at c and 3500 BC. The reconstruction of the beach barrier is hypothetical since it was probably not closed completely, at least at 4100 BC. Occupation occurred on isolated dunes in the former beach plain, mostly oriented in NE-SW direction, which formed 5 to 10% of the landscape. They were relatively flat and stood c. 1-2 metres above the surrounding flats. There are also dunes that were used but not occupied. During the early part of the fourth millennium BC, a tidal channel was present that ran in west-east direction north of Ypenburg. Consequently, tidal influence is more expressed in the northern part of the region. South of the region, the river Meuse had an estuary, resulting in prevalent brackish conditions in the south. In eastern direction, freshwater lagoonal marshes were present (Van der Woude 1983). As a result, a variety of ecotopes was present in and near the region Ypenburg The palaeogeography of Ypenburg was first reconstructed in detail by Cleveringa (2000). The subsurface at Ypenburg consists of channel deposits, wash-over deposits, beach ridge deposits (sand), beach plain deposits (sand) and marsh deposits (peat). Ypenburg is located on a series of dunes that were formed on beach ridges and the beach plain from 4000 BC onwards. The largest dune measures c. 750 x 100 metres, while others have a length of c. 250 metres. The height of the top of the largest dune was 2 metres above the former high water level. A tidal channel dissecting the beach plain became inactive before or during occupation ( BC). The distance to the coast was c. 2.5 km. The surrounding landscape was a moist to wet salt marsh (lagoon fringe) where marine influence occurred during initial occupation. The salt marsh and dunes became less influenced by the sea through time. 73

2 a 0 5 km b km sea beach dune salt marsh reed marsh shrub vegetation alder carr present coastline Figure 3.1 The coastal region, palaeogeographical reconstruction for a) 4100 BC and b) 3500 BC (after Louwe Kooijmans in press). The Meuse estuary is shown in the lower part of the figure. 1 = Wateringen 4, 2 = Wateringse Veld, 3 = Schipluiden, 4 = Sion, 5 = Rijswijk-A4, 6 = Ypenburg, 7 = Schipluiden-Zuidkade, 8 = Ypenburg-Postenkade, 9 = Nootdorp-Driemanspolder. 74

3 3.1.3 Wateringen 4 The subsurface at Wateringen consists of back-barrier deposits (sand and clay, deposited behind the beach barriers). The site was located on a relatively small dune. A second dune with a smaller surface, located next to the site, did not reveal indications of occupation during a coring campaign, and further information on this dune is not available. Peat growth started at c BC. Erosion of unknown scale disturbed the eastern side of the dune at c BC (Raemaekers et al. 1997, 146). The site was located in the southern part of the region where the beach barriers formed a less closed continuum than in the northern part of the region. The reconstruction of the natural vegetation however shows that this southern location did not result in a stronger marine influence during occupation than at other sites Schipluiden (AHR-39) The subsurface at Schipluiden consists of sandy beach plain deposits representing the first phase of coastal progradation, tidal back-barrier deposits (silty clay with sand), and wash-over deposits. Dunes developed on top of the back-barrier and wash-over deposits after 4000 BC. During occupation, the dune was located c. 2.5 km away from the coast. During phase 1, clay was deposited on tidal flats and in tidal channels around the dune. The environment consisted of a salt marsh with both brackish and fresh water influxes. The size of the dune was c. 0.5 ha, and the height of the dune was c. 1.5 metres. During phase 2a, there was no sedimentation of clay or peat. During phase 2b, a reed marsh developed. The height of the dune was c. 0.8 metres. During phase 3, peat growth continued. The size of the dune had decreased to 100 x 30 metres, and the height was c. 0.4 metres. Much of the top of the dune has been eroded. Some preserved parts showed that a podsolic soil developed on top of the dune before the period of occupation Sion (AHR-42) The geology and palaeogeography of Sion is comparable with Schipluiden, since the site was located on a dune 400 metres to the northeast of the site Schipluiden. The site Rijswijk-A4 is located to the northeast of Sion, and one of the four locations at Rijswijk-A4 may be located on the same dune as Sion Rijswijk-A4 The sediments of Rijswijk-A4 were studied by Van der Valk (1992, 1996). The subsurface at the four locations of Rijswijk-A4 consists of tidal or estuarine deposits (sand) and beach plain deposits (sand and clay) including wash-over deposits. Dunes developed on top of the beach plain before BC. The site includes various sub-sites that are presumably located on two dunes, with the tops at 3.45 and c m -NAP Wateringse Veld Wateringse Veld represents four dunes of which the geology is comparable with Ypenburg and Rijswijk-A4. On top of back-barrier deposits, a beach-ridge environment developed. The dunes, measure metres in width, formed at the highest ridges of a beach. The beach-ridge and dunes were covered with peat and clay. Prospective research indicated that Wateringse Veld was not occupied (Bakker and Burnier 1997), but presence of some charcoal suggests that people used the dunes in some other way (see paragraph ). Archaeobotanical data of Wateringse Veld are not available. 75

4 3.2 Nootdorp-Driemanspolder Nootdorp-Driemanspolder is a non-archaeological location from which a pollen diagram is available. The site is located 5 km to the northeast of Ypenburg (coordinates / ). The data of Nootdorp were kindly made available by Alterra, the organisation that was responsible for the archive of K. Koelbloed, former employee of the Stichting voor Bodemkartering. The pollen analysis was performed in 1956 by three persons (K. Koelbloed, JMK and CvdW), possibly with the aim to date the diagram and to obtain an impression of the development of the landscape, but probably not with the aim to reconstruct the vegetation in detail. 1 In order to enable comparison with other diagrams from the coastal region, the diagram has been recalculated based on a total pollen sum, resulting in a pollen sum varying between 171 and 541 pollen grains. The depth of the diagram corresponds with 7.00 to 5.40 m -NAP, assuming that the depth given in the pollen diagram is the depth below surface (which is 4.7 m -NAP). The diagram has been dated but the available sources give contradictive information. The manuscript of the Stichting voor Bodemkartering mentions that two samples were selected for dating, collected at and m below surface. The manuscript refers to GrN-1119: 5600 ± 70 BP ( BC), sampled at m -NAP, which may correspond with m below surface, although the depth ranges do not correspond with each other. The official publication from the responsible institute in Groningen (De Vries and Waterbolk 1958) moreover gives another date: GrN-1119: 5360 ± 70 BP ( BC). Calibration of both dates gives different results, but it nevertheless seems probable that the upper part of the diagram dates to the period before the known Middle Neolithic occupation in the region. Figure 3.2 shows the pollen diagram of Nootdorp. The diagram mainly shows stable, continuous curves, suggesting continuity in the sedimentation and the development of the vegetation. The lower half of the diagram shows evidence of a local presence of Poaceae, Chenopodiaceae and Asteraceae tubuliflorae, suggesting the presence of tidal flats and salt marshes. A single grain of Cerealia-type is present, but this has little meaning without an archaeological context, more dates, or further indications of human impact on the vegetation. The diagram furthermore suggests that much of the material may represent non-local pollen. In the upper part of the diagram comparable indicators of the presence of tidal flats and salt marshes can be observed as in the lower part, now combined with increased values of Artemisia sp. 1 It is unknown to the author whether the diagram has been published before. 76

5 70 90 Pinus Abies Picea Tilia 14C Dates (yrs BP) Depth (cm below surface) Lithology Quercus Betula Ulmus Fraxinus Fagus Upland taxa Corylus Hedera Cerealia Polypodium Botrychium Alnus Salix Myrica Wetland taxa Typha angustifolia-type Filipendula Typha latifolia-type Ericaceae Sphagnum Potamogeton Pediastrum 5360 ± 70? Artemisia Apiaceae Equisetum Rumex Ecologically indeterminate Ranunculus-type Filices Triletae Poaceae Cyperaceae Chenopodiaceae Salt marsh taxa Asteraceae tubuliflorae Armeria Hystrichosphaeridae Indet. Pollen sum Clay Stichting voor Bodemkartering, 1956 Peat Peaty sediment Figure 3.2 Nootdorp-Driemanspolder, pollen diagram based on a total pollen sum, exaggeration 5 x (Stichting voor Bodemkartering) 77

6 3.3 Schipluiden-Zuidkade Schipluiden-Zuidkade (coordinates / ) is a non-archaeological location, 5 km south of the archaeological site Schipluiden, from which a dated pollen diagram is available. The data originate from the State Geological Survey and are published in Hartman (1968), where the location is called Schipluiden. This publication includes only the part of the diagram of Hartman that corresponds with the period before and during the Middle Neolithic occupation in the coastal region. For comparison with other diagrams from the region, the diagram was recalculated based on a total pollen sum, varying between c. 100 and 1200 pollen grains. Table 3.1 shows the lithostratigraphy of the core. Table 3.2 shows the relevant dates, corresponding with the Atlantic and Sub-Boreal. Figure 3.3 shows the pollen diagram. Hartman discusses that the major part of the pollen diagram as published here unto 5.25 m -NAP depth (m -NAP) sediment Phragmites peat amorphous peat with much charcoal clay with marine diatoms peat clayey peat brown clay gyttja peat with much charcoal clay Table 3.1 Schipluiden-Zuidkade, lithostratigraphy (Hartman 1968). mainly represents non-local pollen rain, transported by river water. Furthermore, he argues that a hiatus may be present at 7.25 m -NAP (Hartman 1968, 35). Taxa that may represent the local vegetation are Chenopodiaceae, Asteraceae tubuliflorae and Poaceae in the middle part of the diagram, and Cyperaceae, Apiaceae, Rubiaceae and Polypodiaceae in the upper part of the diagram. The diagram therefore probably reflects the local development of salt marsh vegetation into sedge peat vegetation. Various herbs may represent local vegetation as well. Alnus sp., Salix sp. and Typha latifolia may represent extra-local vegetation. Human impact, if present, would be expected in the upper half of the diagram that corresponds with the Middle Neolithic occupation in the region. It is hardly possible to reconstruct human impact due to the assumed absence of local woodland vegetation, while the visible changes in the curves of herbs cannot be related to human activity with certainty. depth (m -NAP) age (yrs BP) age (yrs cal BC, 2σ) dated material ± 60 BP 3340 BC (95.4%) 2910 BC amorphous peat ± 60 BP 3670 BC (78.4%) 3490 BC clay 3460 BC (17%) 3370 BC ± 45 BP 3960 BC (95.4%) 3710 BC peat ± 70 BP 5210 BC (2.8%) 5160 BC peat 5080 BC (92.6%) 4720 BC Table 3.2 Schipluiden-Zuidkade, 14 C dates (Hartman 1968). 78

7 ± ± 60 Lithology 14C Dates (yrs BP) Depth (cm - NAP) Pinus Picea Tilia Quercus Ulmus Upland trees and shrubs Betula Fraxinus Carpinus Fagus Populus Taxus Ilex Acer Hedera Viscum Corylus Viburnum Hippophae Juniperus Rosaceae Artemisia Valerianella Polygonum aviculare Upland herbs Spergularia Verbascum-type Papaver Plantago media Polypodium vulgare ± ± Alnus Salix Myrica Rhamnus frangula Typha angustifolia Lysimachia vulgaris Caltha Typha latifolia Alismataceae Wetland trees, herbs and spore plants Valeriana officinalis Filipendula Rumex acetosella Lotus Sagittaria Rumex maritimus Solanum dulcamara Thalictrum Hydrocotyle Hottonia Butomus Lythrum Polygonum amphibium Iris Ericaceae Empetrum Sphagnum Nymphaea Isoetes Potamogeton Open water Chenopodiaceae Salt marsh Asteraceae tubuliflorae Triglochin Plumbaginaceae Plantago coroponus Glaux Peat Clay Figure 3.3 part 1. 79

8 Poaceae Cyperaceae Apiaceae Rubiaceae Brassicaceae Caryophyllaceae Ranunculus Ecologically indeterminate Asteraceae liguliflorae Malvaceae Vicia Euphorbia Lamiaceae Polypodiaceae Indet. Pollen sum Total pollen concentration (n/mm 3 ) National Geological Survey, 1966 Figure 3.3 Schipluiden-Zuidkade, pollen diagram based on a total pollen sum, exaggeration 5 x (based on Hartman 1968), part Ypenburg-Postenkade Ypenburg-Postenkade referes to the location of a pollen core (coordinates / ) that was sampled directly south of the dune where the site Rijswijk-Ypenburg is located. The discussion of the data is based on the diagram presented in Kooistra (2004). Samples were collected with pollen boxes at m -NAP, and subjected to pollen and macroremains analysis. The pollen sum consisted of 600 pollen grains based on a total pollen sum excluding taxa of water plants and non-pollen palynomorphs. The resulting diagram dates to the period between c BC. Only the three lowest spectra (4.44, 4.37 and 4.36 m -NAP), dating to the period before BC (GrN-25885: 4420 ± 90 BP), possibly correspond with the Middle Neolithic occupation in the region. The sediment of the spectra consisted of clay and clayey peat. The arboreal pollen percentage of 20-25% indicates an open landscape where woodland was scarce or absent. The herb pollen indicates the presence of salt marshes and eutrophic marshes. The sample at 4.36 m -NAP contained pollen of Cerealia-type, Urtica dioica, Artemisia sp. and Plantago lanceolata, which can be considered as anthropogenic indicators. The presence of these taxa may relate to Middle Neolithic occupation, but this is not demonstrated due to their scarce presence and absence of further archaeological indicators in the core. The spectra above those that potentially correspond with the Middle Neolithic occupation in the region represent reed marsh vegetation with reed, sedges and ferns. These younger spectra do not show explicit evidence of human impact (Kooistra 2004). 80

9 3.5 Rijswijk-Ypenburg Archaeology The discussion of the site of Ypenburg only includes preliminary results since only general results were published at the time of this research (Koot and Van der Have 2001). 2 The site (coordinates / ) was excavated on a large scale, covering several hectares. The site was occupied in the period BC, corresponding with the Hazendonk group. Two main occupation phases have been distinguished, 3/C and 11/K. There were several other phases, including an early phase 2/B. Important features are four two-aisled houses (maximal 4.5 x 10 metres), unlined wells and a cemetrey with men, women and children. The preliminary results indicate that the site functioned as a year-round occupied settlement during some occupation phases. See the final site report (Koot et al. 2008) for further information. Preliminary results on botanical macroremains were published in Van Haaster (2001) and Koot and Van der Have (2001). Hazenberg Archaeology and BIAX Consult kindly gave access to unpublished information on pollen, macroremains, coprolites, tubers, processed carbonised plant remains, wood and charcoal that is part of the final site report. The text below is based on Van Beurden (2008a, b), Kooistra (2008b), Kooistra and Hänninen (2008), Kubiak-Martens (2008), and Van Waijjen (2008), all published in Koot et al. (2008) Pollen analysis The pollen analysis comprised one sample of phase 2/B that took place just before phase 3/C, three samples from phase 3/C and three samples from phase 11/K. All samples were collected from a single section at the edge of the dune. 3 The analysis was based on a total pollen sum that varied between 400 and 1400 pollen grains, usually reaching a value of max. 650 pollen grains. The sample of phase 2/B indicates the local presence of freshwater marsh vegetation without alder, and the extra-local presence of salt marsh vegetation or influxes of brackish water. The samples of phase 3/C are characterised by an increase in taxa that tolerate brackish conditions, although freshwater taxa were still present in the extra-local vegetation. During phase 11/K, the importance of taxa that tolerate brackish conditions decreased, while the importance of Alnus sp. and Poaceae increased, indicating that the environment became less affected by marine influence and became a freshwater environment instead. In the last spectrum, Dryopteris-type increases strongly. Reconstruction of the vegetation on the dune is difficult for all phases. Pollen of dune shrub vegetation is present in small quantities (Acer campestre-type, Hippophae rhamnoides and Juniperus communis) and probably represents the dune vegetation (Van Beurden 2008b). Indications of human impact in the pollen diagram are minimal because of the character of the natural vegetation and because of the small number of samples. The location of the sample point may play a role as well. The importance of shrubs decreased at the end of phases 3/C, which may be related to human impact or may be a statistical result from the increased percentages of Cyperaceae that were present in the local vegetation. The most direct indications of human impact are the presence of pollen grains of Cerealia-type in all phases and the presence of grains of Hordeum-type in phase 3/C. Taxa indicative of ruderal, disturbed terrain show relatively high values at the end of phase 3/C and in 11/K, corresponding with the presence of Cerealia-type pollen (Van Beurden 2008b). Analysed coprolites are interpreted as coprolites from carnivores such as dog, fox and cat. A coprolite from phase 3/C was relatively rich in pollen of Corylus sp. and Alnus sp., which suggests the presence of these taxa in the surroundings of the site (Van Waijjen 2008). 2 The data on Ypenburg presented in this chapter are nevertheless more detailed than in the chapters on gathered food plants and weeds since this chapter was finished in a later stage. The results by Koot and colleagues were published when the research for this study on the Dutch Late Mesolithic and Early and Middle Neolithic wetland sites was finished but not published. 3 The location of the pollen core was not known to the author during the writing of this study since the site report was not published yet. 81

10 3.5.3 Macroremains analysis The analysis of macroremains is based on 21 samples of varying volumes 4 sieved on a 0.5 mm sieve and c samples of 5 litres sieved on a 2 mm sieve. Some of the 21 samples were selected because of the context (refuse layers, hearths and pits), diversity, richness, and presence of food plants. The selection comprised both rich and poor samples (Kooistra 2001). The resulting assemblage contained taxa of dune shrub vegetation, some herbs indicative of shade, taxa indicative of dry to moist, disturbed terrain, and taxa of freshwater marshes and salt marshes. Taxa of open water were only scarcely found in unlined wells. Taxa of salt marsh vegetation were interestingly found in unlined wells as well. The numbers of samples of the various occupation phases do not allow analysis of changes through time (Van Beurden 2008a). Carbonised finds of probable food plants included Corylus avellana, Prunus spinosa and Rosa sp. Another potential food plant of which seeds were found in a carbonised state is Asparagus officinalis (although it may have been the shoots that were consumed). Waterlogged finds of probable food plants included Malus sylvestris, Pyrus communis ssp. pyraster, Crataegus monogyna, Sambucus nigra, Prunus padus, Berberis vulgaris, Rubus caesius, Rubus fruticosus, Rubus idaeus and Physalis alkekengi. Other edible taxa may have been consumed as well. The group of probable food plants from Ypenburg contains some taxa that are not usually found at comparable sites, such as Pyrus communis ssp. pyraster, Berberis vulgaris, Physalis alkekengi and Asparagus officinalis (Van Beurden 2008a). Contextual indications of use are available for Ruppia maritima, since carbonised fruits of this species were found in a hearth (Van Beurden 2008a). Analysis of carbonised tubers and carbonised processed food remains resulted in identifications of probable food plants as well. Analysis of carbonised parenchyma resulted in identification of Allium sp. (possibly A. scorodoprasum or A. vineale), Typha sp. (possibly T. angustifolia) and Pteridophyta (ferns). These plant remains are indeed known as human food sources and are also know from other excavations in Northwestern Europe (see chapter 9). Analysis of carbonised processed plant remains collected from pottery sherds resulted in identification of the epidermis of emmer wheat present in a sample, together with stem or leaf remains from plants, animal meat or fat, and fats from plants (Kubiak-Martens 2008). Remains of crop plants found at Ypenburg are carbonised grains of emmer wheat (Triticum dicoccon) and naked barley (Hordeum vulgare var. nudum), and carbonised chaff of emmer wheat. There were some grains that showed some similarity with einkorn (Triticum monococcum) that were interpreted as grains from the top of emmer ears (Van Beurden 2008a, 316). There were additionally some grains that showed some similarity with hulled barley (Hordeum vulgare var. vulgare) that were interpreted as deformed grains of naked barley. Emmer chaff was found in five samples only. The low frequency of chaff and absence of chaff of naked barley is probably caused by the recovery methodology, since the majority of samples were sieved on a 2 mm sieve. The ratio of the two cereals, based on the number of grains, suggests general dominance of emmer wheat, but the validity of this conclusion is limited since many cereal grains could not be identified up to genus level and the ratio may have shifted through time. Identified stem remains of Poaceae may represent stems of cereals (Van Beurden 2008a). The site revealed a varied list of indicators of disturbed and ruderal terrain, possibly representing arable weeds (Van Beurden 2008a; see also paragraph ). Taxa that were found in a carbonised state are cf. Bromus sp., Chenopodium album, Galium aparine, Malva neglecta, Persicaria lapathifolia, Rumex sp., Stellaria media, Vicia sp. and Vicia sativa. Two samples rich in emmer chaff remains did not contain any carbonised macroremains of potential weeds. Chenopodiaceae, cf. Bromus sp., cf. Galium sp. and Malva neglecta were each found in a sample with few cereal remains without mixture of other taxa, which suggests 4 The samples published in Van Haaster (2001) had a volume varying between 1 and 5 litres. 82

11 that these taxa may represent arable weeds. The earlier archaeobotanical report also mentions a find of a single waterlogged grain of cf. Avena sp., interpreted as a possible find of A. fatua, a weed in cereal fields (Van Haaster 2001). Other finds of Avena sp. at the sites studied only includes pollen identifications at Urk-E4 (Van Smeerdijk 2001) Wood and charcoal analysis The wood remains, collected from the refuse layers, features and residues from the 2 mm sieve, were grouped into wood with no or scarce traces of working (N = 243) and artefacts (N = 15). Most of the presented wood corresponds with the occupation phases. The wood contained 65 unidentified remains. Table 3.3 shows the taxa identified in the assemblages of unworked and worked waterlogged wood, and charcoal. The unworked wood assemblage was dominated by Alnus sp., followed by Pinus sp., Quercus sp., Salix sp. and Fraxinus excelsior. This frequency appears to be constant through time. Phase 11/K contained only wood of Alnus sp., Salix sp. and Cornus sp.; the small number of taxa is probably related to the small number of wood remains found from this phase. The presence of Pinus wood is remarkable in view of the absence of this species at other sites in the region and the absence of macroremains. taxon category unworked wood worked wood charcoal taxon category unworked wood worked wood charcoal Acer sp Prunus padus/spinosa Alnus sp Prunus spinosa Betula sp Prunus-type Cornus sp Quercus sp Corylus avellana Rhamnus cathartica Euonymus europaeus Rhamnus frangula Fraxinus excelsior Salix sp Ilex aquifolium Sorbus sp Juniperus communis Taxus baccata Ligustrum sp. - - cf. + Ulmus sp Pinus sp Viburnum opulus Pomoideae Viscum sp. - - cf. + Prunus avium/padus = present - = not present Table 3.3 Ypenburg, unworked wood, worked wood, and charcoal (Kooistra 2008b; Kooistra and Hänninen 2008). 83

12 One piece of Pinus wood is dated to BC (9110 ± 50 BP). In addition, artefacts of Pinus wood are absent despite frequent presence of unworked Pinus wood at the site, which may be related to the poor quality of the old wood. It is therefore suggested that the wood of Pinus sp. represents reworked material from the Boreal (Kooistra and Hänninen 2008). Parallels are known from several Dutch Neolithic sites (see chapters 6 and 7), but not from other sites in the coastal region. The 15 artefacts comprised posts, possible planks, pointed roundwood, hafts, wickerwork and an unidentified artefact. Posts and possible posts were made of Alnus sp., Quercus sp., Salix sp. and Fraxinus excelsior. Two planks were made of Alnus sp. Pointed roundwood was made of Juniperus communis, F. excelsior, Alnus sp. and Rhamnus cathartica. Hafts were made of Quercus sp. and Salix sp. The haft of Quercus sp. was short (14 cm in total) and is interpreted as a possible haft of a sickle. The wickerwork was made of Cornus sp. (C. sanguinea). It is tentatively interpreted as a preparation for a fish trap (Kooistra 2008b), which corresponds with other selective use of red dogwood for fish traps (Out 2008b). The excavation also revealed two pieces of rope. All taxa that were used for artefacts were relatively common in the wood and charcoal assemblage (Kooistra 2008b; Kooistra and Hänninen 2008). Charcoal (N = 1553) from the occupation phases 3/C (c. 66%) and 11/K (c. 33%) was collected from sieve residue samples of the 2 mm sieve and from features. The charcoal assemblage of phase 3/C was dominated by Alnus sp., Quercus sp. and Fraxinus excelsior. Identification of the charcoal from one of the two hearths that were both dated to phase 3/C resulted in four identifications only: Salix sp., Alnus sp./betula sp., Alnus sp. and Pinus sp. Analysis of the second hearth showed use of Quercus sp., Corylus avellana and Alnus sp. These results do not indicate selection of fuel based on the qualities of the wood (Kooistra and Hänninen 2008). The assemblage of phase 11/K contained relatively more Quercus sp., Alnus sp., Salix sp., Corylus avellana and Sorbus aucuparia than the assemblage of phase 3/C. Specific contexts at Ypenburg resulted in identification of Quercus sp., Salix sp. and cf. Pomoideae. These contexts may represent selective use of wood based on the quality of the wood, but the contexts are not understood which makes interpretation difficult. The charcoal indicates use of primarily fresh and dry wood for fuel, and additionally of moist wood and brushwood affected by fungi (Kooistra and Hänninen 2008) Discussion The natural vegetation on the dune at Ypenburg probably consisted of scarce dune shrub vegetation, while the marshes that surrounded the dune developed from freshwater marshes to salt marshes and finally into freshwater marshes again during occupation. The analysis of waterlogged wood and charcoal revealed an unexpected large variety of taxa. The data indicate that the exploitation area of the site included dune shrub vegetation, alder carr and deciduous woodland vegetation. Kooistra and Hänninen (2008) conclude that especially the finds of Acer sp., Fraxinus excelsior, Ilex aquifolium, Quercus sp., Taxus baccata and Ulmus sp. indicate the presence of moist eutrophic woodland with changing water tables. The precise location of such vegetation in the region is not known. 5 The macroremains assemblage from Ypenburg contains four taxa that are unique for the sites of the Hazendonk group in the coastal region as well as for other Late Mesolithic and Early and Middle Neolithic Dutch wetland sites: Asparagus officinalis, Pyrus communis ssp. pyraster, Physalis alkekengi and Berberis vulgaris. There is a chance that research methods play a role here, since the macroremains samples were sieved with water from a nearby ditch and since the samples were stored uncovered in open air for an unknown period. The analysts were however aware of these sources of possible contamination, and the four unique taxa in the macroremains assemblage are not taxa that were found in large numbers in the sieving water, and are not 5 Palaeogeographical reconstructions do not give detailed information since part of the region was destroyed by erosive channel activity. 84

13 transported by wind through the air either. An alternative explanation for their presence is that the large number of samples investigated plays a role (3000 samples of 5 litres sieved on a 2 mm sieve). Finally, the large number of new taxa may possibly be related to the site function and agency (human choice). There are no indicators of the import of taxa since they could all have been part of the natural vegetation in the coastal region. Local crop cultivation may have occurred at Ypenburg since the size of the dune (7.5 hectare) would have enabled local cultivation and since the environment in the exploitation area was probably suitable for arable farming, at least at those locations that were hardly affected by flooding during the major part of the year. The botanical finds of crop plants included pollen and macroremains of emmer wheat and naked barley. Chaff remains of emmer were found, but there is no information on the presence or absence of chaff remains of naked barley. There are indications of burning of the vegetation at the dune and disturbance of the soil (Kooistra et al. 2002), which are possibly related to arable farming (further discussed in paragraph and chapter 11). Use-wear analysis demonstrated the presence of flint with cereal gloss running in a longitudinal direction, which forms a major argument in favour of local arable farming (Van Gijn et al. 2006, 154). The wooden artefacts show the presence of a possible sickle haft. The available data overall provide positive indicators in favour of arable farming. More archaeological interpretations of the site on site function, mobility system, etc. are however necessary in order to make a final conclusion. 3.6 Schipluiden Archaeology The coastal dune site of Schipluiden-Harnaschpolder (coordinates / ) was almost fully excavated in 2003 by Archol B.V. in cooperation with Leiden University. The results are published in Louwe Kooijmans and Jongste (2006). The Holocene stratigraphy on the slopes of the dune allowed the distinction of stratigraphical units that could be dated. The settlement history was divided into four phases: phase 1 at c BC, phases 2a and 2b at c BC and phase 3 at c BC. The find assemblage of phase 1 is relatively small while the assemblage of phase 2 is rich. The pottery and the 14 C dates indicate that occupants at Schipluiden were part of the Hazendonk group. The features comprise a large number of unlined wells, concentrated at the northern slope, pits, hearth pits, some burials, a deposition pit, ditches, fences, and clusters of many postholes interpreted as four or five house sites or yards. The finds comprise flint, stone, pottery, ornaments and organic remains. The concentration of the posts and the distribution patterns of the stratified archaeological remains indicate the continuous presence of four households or farmyards, certainly from phase 2a, and possibly already from phase 1 onwards. The fences were present around the site from phase 2b onwards. The average group size of these four households is estimated at 25 persons (Louwe Kooijmans 2006b). The pottery is characteristic of the Hazendonk group (Raemaekers and Rooke 2006). The flint assemblage consisted of a large component of rolled pebbles and a small component of high quality imported material. The source of the rolled pebbles is expected to be located relatively close to the site (at the Belgian or Dutch coast). The imported flint indicates that the contact area of the site was in a southern and southeastern direction. Use-wear analysis of the flint resulted in indications of various activities including wood working, plant processing and cereal harvesting. In contrast to other Dutch Mesolithic and Neolithic sites there are no indications of tools that were used transversely for the processing of siliceous plant, indicating a change in techniques or a change in subsistence. Phytolith analysis of the querns indicated cereal processing. Use-wear analysis of bone and antler was indicative of plant processing (sewing), wood working and hide working (Van Gijn et al. 2006; Van Gijn and Houkes 2006). 85

14 The extended broad-spectrum economy of the site at Schipluiden was based on animal husbandry, crop cultivation, hunting, fishing, fowling and gathering. The ratio of domestic, terrestrial (wild) and aquatic protein food sources is estimated at 2:1:3 respectively. The site is considered to be occupied by complete households continuously and year-round at least from phase 2a onwards, based on the presence of houses, the skeletal remains of men, children and possibly a female, the local production of pottery, the broad range of artefacts and the wide diversity of materials, the production and repair of artefacts, the broad range of activities as indicated by the use-wear analysis, the diversity of food sources, the indications of seasonality and the composition of the wood assemblage (Louwe Kooijmans 2006b). The zoological and botanical data support the archaeological arguments for permanency (Kubiak-Martens 2006a; Zeiler 2006a, b). The animal bone assemblage was dominated by cattle, wild boar and red deer. The importance of cattle slightly decreased through time while the importance of hunted mammals increased. The domestic animals present at the site are dog, cattle and pig, while sheep and goat were absent. There are no indications of the use of dairy products (Boon 2006; Brinkhuizen 2006; Zeiler 2006a, b) Diatom and pollen analysis The results of the diatom analysis are based on De Wolf and Cleveringa (2006). Diatoms indicate that at the start of occupation the site was located in a dynamic estuarine environment, representing the transition from saline to freshwater conditions. Freshwater conditions prevailed during the later occupation phases. The flooding frequency decreased through time from regular high tides into irregular floods during spring tides and storms, into highly irregular floods (De Wolf and Cleveringa 2006). The discussion of the pollen analysis is based on Bakels (2006). Three sections were sampled with sample boxes at the northwestern and southeastern edges of the dune. The sections were combined with samples from four unlined wells located at the northwestern side of the dune and from eight coprolites. The analysis is based on a volume of 1 cm 3 and based on a total pollen sum. The pollen sum of the sections varied between 869 and 2486 pollen grains. The number of spectra counted per section varies between four and eight. The coprolites and half of the wells were poor in pollen. The pollen analysis indicates the transition from salt marsh present all around the site into reed marsh vegetation during occupation, followed by sedge marsh vegetation. Taxa that initially dominate the pollen assemblage are Chenopodiaceae, Asteraceae, and a variety of taxa that possibly represent salt marsh taxa. Later Poaceae and afterwards Cyperaceae dominate in combination with taxa of eutrophic freshwater marshes. The results of the three sections are very similar. There is evidence of the absence of woodland vegetation on the dune since the percentage of pollen of trees is very low and since the percentage decreases at the transition from clay to peat, which indicates that the pollen found in the clay represent secondary transported pollen. Dune shrub vegetation was possibly present at the site, since the diagrams occasionally contain pollen of Prunus avium/spinosa, Cornus sanguinea, Viburnum opulus and Hippophae rhamnoides (all taxa that produce little pollen that is not spread over large distances). The shrub pollen is best represented in the stratigraphical units that correspond with occupation phase 1. The unlined wells, dating to phase 1-2a, do not however contain pollen of shrubs, which may be related to their specific location or with the low pollen sum of the well samples. The pollen diagrams from the sections theoretically provide information on human impact since they correspond with occupation phases 1, 2a/2b/3 and the period after occupation. The low number of analysed spectra and the open vegetation however hamper a detailed analysis of human impact on the vegetation. The fact that shrub pollen is mostly found in the sediments from phase 1 indicates that occupation resulted in a decreased presence of shrub vegetation. Pollen identifications of possible anthropogenic indicators are Cerealiatype, Plantago lanceolata, Plantago media, Artemisia sp., Polygonum aviculare and Chenopodiaceae. 86

15 3.6.3 Macroremains analysis The discussion of the macroremains is based on Kubiak-Martens (2006a). For analysis of botanical macroremains three types of samples were collected: 5-litre samples from features, (wells, pits, hearths and postholes), 5-litre interval samples collected every six metres, remains collected from the 4 mm sieves, and handpicked remains (N = 500). After assessment of the samples, 60 samples of the first two types of samples with a volume of 0.5 litres were analysed after sieving on a 0.25 mm sieve. The samples mainly represent stratigraphical units and wells, and additionally other features. The samples from pits, hearths and postholes could not be attributed to a single phase. Although the aim of the sampling program and the analysis was to obtain data of similar contexts from all occupation phases, the data set is influenced by taphonomy and preservation. The interval samples collected from units from the early phases (1/2a) mainly contained waterlogged remains, while in contrast similar samples from phases 2 and 3 contained mainly carbonised remains. The samples from most features also contained mainly carbonised remains, except from samples from wells that contained mainly waterlogged remains. Remains from phases 1/2a were mainly retrieved from the northwestern side of the dune while remains of later phases were mainly retrieved from the southeastern side of the dune. Comparative analysis of the remains through time and space is therefore possible on a restricted scale only. The macroremains indicate the presence of a variety of biotopes in the local and extra-local surroundings of the site: high salt marshes, eutrophic freshwater marshes that may have been somewhat brackish, dry grasslands, moist eutrophic grasslands, terrain rich in nitrate, disturbed and/or trodden terrain, shrubs, alder carr and woodland. There are a few taxa indicative of open water. The variation of taxa and their ecological range within single samples is generally very high. It appears that most interval samples and samples from features do not represent assemblages deposited during a single deposition process but represent settlement refuse instead. The general presence of salt marsh taxa in the macroremains assemblage can be explained in various ways. Kubiak-Martens (2006a, 319, 329) argues that these macroremains entered the site together with animal dung after the grazing of domestic animals on the marshes, as gathered animal fodder, and together with the cereals, i.e. as arable weeds (further discussed in paragraph ). Natural deposition processes should however be taken into consideration as well. Tidal flats and salt marshes were present around the dune during phase 1 and possibly during phase 2a. Macroremains of salt marsh taxa may have been deposited with the drift litter during all phases. This drift litter may have been used as fuel or may have been added to the fire for other reasons (see paragraph 5.5.3), resulting in carbonisation. Furthermore, for Wateringen 4 it has been suggested that the presence of Suaeda maritima, a salt marsh species, may be related to import of clay to the site (Raemaekers et al. 1997), and this could explain presence of salt marsh taxa at Schipluiden as well. The taxa that are found in a carbonised state at Schipluiden are provided in Kubiak-Martens (2006a) and in the regional synthesis below (paragraph ). The list of taxa is long and contains taxa of all distinguished ecological groups. The potential food plants most frequently found in carbonised and waterlogged state are Prunus spinosa and Malus sylvestris. One of the wells contained a concentration of waterlogged fruits of Prunus spinosa, supporting intentional collection by people. It is argued that Prunus spinosa and Malus sylvestris were present on the dune (Kubiak-Martens 2006a), although the taxa are probably overrepresented due to intentional gathering from other locations. The distribution of the macroremains of Prunus spinosa strongly corresponds with the distribution of other find categories, which indicates that the distribution of Prunus spinosa is indeed related to human activities (cf. Kubiak 2006a, 329). Interestingly, the botanical macroremains contained carbonised remains of tubers of Beta vulgaris ssp. maritima (sea beet), Allium sp. (onion/leek) and Bolboschoenus maritimus (sea club-rush). Food crusts from pottery were analysed as well, containing fruits of Atriplex sp./suaeda sp. (Kubiak-Martens 2006b). 87

16 The presence of the woodland taxa Carex elongata, Moehringia trinervia, Stachys cf. sylvatica, Carex remota, Circaea lutetiana and Fallopia dumetorum all indicate the presence of somewhat shaded terrain (not necessary completely shaded). The frequency of most of these taxa is 3.5% in a waterlogged state, they do not occur in a carbonised state and they mainly occur in samples form phases 1, 1/2a and 2 (not phases 2b and 3). Their waterlogged state indicates that the taxa may indeed represent the natural vegetation at the dune, while their low frequency and decreasing presence supports the results of the pollen that the dune was very open and that it became more open through time. In addition to shrub vegetation, the houses and possible other structures may have created some shady patches at the dune. Moehringia trinervia is the only species of this group occurring frequently in a waterlogged state (frequency 33%) and occurring in a carbonised state. M. trinervia prefers partly shaded woodland and shrub vegetation on slightly moist sand ground rich in humus, where the decomposition rate of organic material is high. These conditions are present if the water table changes strongly, where the soil is rich in lime (calcium) or where vegetation is cleared. It occurs frequently under dune shrubs, along small channels and at vegetation clearances (Weeda et al. 1985). The high frequency of the species could indicate frequent occurrence of these conditions, or could alternatively indicate that people used the species in some way Crop plants The macroremains assemblage contained mainly carbonised grains and chaff remains of Triticum dicoccon and Hordeum vulgare var. nudum that were generally found together in samples. The chaff remains of naked barley consisted of rachis internodes and rachis fragments (rachis fragments: a part of the rachis consisting of several internodes still attached to each other; quantity unknown). The identifications additionally included many stem fragments of Poaceae (in carbonised and waterlogged states), possibly representing Cerealia. The waterlogged remains contained only scarce chaff remains of Triticum dicoccon. The data show a trend that naked barley dominates the crop assemblage in the first phases (phases 1 and 1/2a), while emmer dominates the assemblage during the later phases, although the number of samples of phase 1 is not representative. The crop plant remains are extensively described in Kubiak-Martens (2006a). Cereal remains were present in the majority of the investigated samples (52 of 60; possibly related to sample selection). The presence of cereal remains in all kinds of samples and the distribution pattern of the cereals (see the publication) indicate that the cereal remains were scattered over large parts of the site and were present in many samples, though in low densities. Most grains and chaff remains were found at the southeastern side of the dune in several clusters that are interpreted as farmyards, which indicates cereal processing and consumption on the household level (Kubiak-Martens 2006a). The number of cereal remains in individual samples is relatively large compared with most other Dutch wetland sites. The number of grains in single samples does however not exceed 50 grains and pure concentrations of cereals seem to be absent, especially since the samples generally contain macroremains of a variety of other taxa as well (potential arable weeds and food plants). There are two samples that may represent emmer chaff concentrations (Kubiak-Martens 2006a) Arable weeds Kubiak-Martens (2006a) suggested that two groups of taxa found in a carbonised state represent the arable weeds of the site Schipluiden. The first group consists of taxa that indicate disturbance and contains Chenopodium album, Galium aparine, Persicaria maculosa, Solanum nigrum, Brassica rapa and Vicia hirsuta. It is argued that the macroremains of these taxa arrived together with the harvest at the site since these taxa were consistently found together with charred grain and chaff remains (Kubiak-Martens 2006a, 329). Other potential arable weeds of this group that are listed are Atriplex patula/prostrata, Capsella bursapastoris, Sisymbrium officinale and Malva sp. The second group consists of taxa that are characteristic of salt marshes and drift litter zones on the salt marshes and contains Althaea officinalis, Apium graveolens, Hordeum 88

17 marinum, Carex distans, Ruppia maritima and Suaeda maritima. It is argued that the frequency of carbonised salt marsh taxa corresponds with the frequency of carbonised remains of crop plants and arable weeds from the first group. It is therefore hypothesised that the fields were located on the salt marshes and possibly on nearby dunes (Kubiak-Martens 2006a, 333). The hypothesis that the fields were located on the salt marshes is supported by evidence from the experiments of Van Zeist (1976) that yielded weed assemblages consisting of a comparable combination of taxa, and by the similar evidence from comparable Dutch sites dating to the Late Neolithic (Kubiak-Martens 2006a, 329). The precise distinction of arable weeds at Schipluiden is more difficult than suggested, since closed context containing pure assemblages resulting from crop processing were scarce. There are three samples that may represent unmixed assemblages resulting from crop processing, since they contain relative large concentrations of chaff remains of emmer wheat (N = 111, N = 144, N = 120) as well as other cereal remains (sample nr. 37, 43 and 47). The samples all contain seeds of Malva sp. and stem fragments of Poaceae that both may represent arable weeds (although the stems may also represent the crops themselves), but the remaining content of the samples varies. The first sample contains few taxa that almost all may represent arable weeds, but the third sample that contains the most potential arable weeds also contains the most taxa that are unlikely to represent arable weeds. This variety suggests contamination of this sample and hampers distinction of the weeds. Some comments can be made on the relationship between the two proposed groups of weeds. The taxon freq. (%) Poaceae, stems 56 Poa sp. 50 Prunus spinosa 50 Hordeum marinum 35 Malva sp. 25 Phragmites australis, stem fragments 23 Galium aparine 23 Chenopodiaceae 19 Galium sp. 19 Althaea officinalis 17 Malus sylvestris, parenchyma 15 Ruppia maritima 15 Solanum nigrum 15 Atriplex patula/prostrata 15 freq. = frequency Table 3.4 Schipluiden, carbonised macroremains, the frequency of the taxa (%) that occur in a high frequency ( 15%) in the 52 samples that contain carbonised cereal remains. distinction of the first group of arable weeds is based on the ecology of the taxa, and it is suggested that the distinction is based on consistent co-occurrence with crop plants as well. Almost all samples however contain not only cereals and taxa that indicate disturbance (typical arable weeds), but also taxa that represent shrubs, freshwater marshes, meadows and/or salt marshes. It must therefore be questioned whether the listed taxa indeed are found together with crop plants consistently. Table 3.4 shows the taxa that occur in 15% or more of the samples with cereal remains. Only carbonised remains are considered in order to assure the relation between the crop plants and the potential weeds. Firstly, this table shows that the data set does not support that the presumed weed taxa of group 1 are found consistently together with carbonised cereal remains. Only Galium aparine and Malva occur in c. 25% of the samples that contain carbonised cereal remains, Atriplex patula/prostrata and Solanum nigrum occur in 15% of the samples while the other taxa of group 1 occur in less than 10% of the samples (together with many other taxa). Secondly, the taxa in table 3.4 represent a combination of food plants, salt marsh taxa, forbs and indicators of disturbance. The broad ecological and functional variety of the taxa indicates that it is unlikely that this complete group of taxa represent arable weeds. Indeed, the association between 89

18 the cereal remains and the other carbonised remains must be doubted since the variety of taxa in the samples indicates that the samples do not represent pure assemblages but instead an assemblage of scattered remains that are the result of various deposition processes. Table 3.4 shows that taxa occurring in a relatively high frequency and with ecological preferences similar to arable weeds are Poa sp., Hordeum marinum, Malva sp., Galium aparine and Chenopodiaceae. Therefore, the analysis of frequent co-occurrence of taxa in a carbonised state together with carbonised cereal remains indicates that these taxa may represent arable weeds. As suggested by Kubiak-Martens, the frequency analysis of carbonised taxa in samples that contain carbonised cereal remains also demonstrates that some of the mentioned salt marsh taxa are regularly present in a carbonised state in samples that contain carbonised cereal remains: Hordeum marinum, Althaea officinalis and Ruppia maritima. However, these taxa do not grow together in a single habitat in the modern-day vegetation (Weeda et al. 1987, 1991, 1994). Ruppia maritima is a water plant and is not expected to have functioned as a crop weed. 6 Some of the other frequently found taxa may represent vegetation of the (high) salt marsh as well (Chenopodiaceae, Poa sp., Galium aparine, Phragmites australis, Solanum nigrum and Atriplex prostrata), but this is unlikely for some of the other taxa (Malva sp., Prunus spinosa and Malus sylvestris). The samples furthermore only contain a few salt marsh taxa. This again indicates that the samples probably represent material of mixed origin. The presence of carbonised salt marsh taxa can moreover also be explained by hypotheses other than cultivation on the salt marsh (discussed above). The new results from the frequency analysis of potential arable weeds can be used for reconstruction of the location of the arable fields, although it is not assured that the most frequent taxa all represent the same context. The analysis indicates that Poa sp., Hordeum marinum, Malva sp., Galium aparine and Chenopodiaceae most likely represent arable weeds. There were probably also other weeds, but these are difficult to distinguish. Hordeum marinum stands out from the list of likely arable weeds since it tolerates considerable brackish conditions. The dominance of the remaining taxa, which are not necessarily indicative of brackish conditions, indicates that in the particular case of Schipluiden it is not probable that crop cultivation took place on the high salt marshes, and that arable farming on the beach plain behind dunes, on dune slopes or on a low dune that was not used for occupation is more likely. Marine influence at such locations near the dune of Schipluiden was presumably considerable during the first occupation phase, as indicated by the geological reconstruction and the pollen analysis, but decreased afterwards (see paragraphs and 3.6.2). Fields located in the dunes would have been closer to the settlement, reducing travel time and reducing the risk of flooding. The location of the fields near the site corresponds with the scarce exploitation of other marine resources. Furthermore, the presence of fields at locations more inland than the high salt marshes corresponds with the dominance of emmer wheat during the majority of the phases. The salinity of the environment at Schipluiden decreased through time. Although emmer wheat can be cultivated on the high salt marshes, naked barley tolerates brackish conditions better than emmer wheat (Bottema et al. 1982). Therefore, it is expected that naked barley would dominate the crop assemblage continuously when the fields would have been located on the salt marshes continuously. This is further discussed in paragraph The taxa Hordeum marinum, Malva sp., Galium aparine and Chenopodiaceae as well as the classical potential arable weeds are mainly annuals. This indicates that the field were not shifted frequently but were probably laid out permanently (see also chapter 10). 6 The fruits of Ruppia maritima may have arrived at the site in one of the ways described in paragraph (in other ways than together with the crops). Alternatively, fruits of Ruppia maritima may have been used intentionally for an unknown purpose, since they were found in a carbonised state in a hearth at Ypenburg (see paragraph 3.5.3), and since a carbonised find is known from Wangels (see paragraph 9.4.3). 90

19 3.6.6 Wood and charcoal analysis The discussion on the wood and charcoal remains is based on Kooistra (2006b) and Louwe Kooijmans and Kooistra (2006). The assemblage of wood consists of unworked and worked wood. The excavated unworked wood was only partially investigated. In the publication of the site, the unworked wood is discussed together with wood remains that show traces of working other than artefacts or posts. This combined assemblage (N = 440), only representative of occupation phase 1-2a and phase 2a, is dominated by Alnus sp., Pomoideae, Prunus sp., Juniperus communis and Salix sp. (see table 3.5). The assemblage further contains shrubs, some deciduous trees and a few identifications of Taxus baccata and Fraxinus excelsior. The unworked wood is assumed to represent the natural vegetation in the exploitation area of the site. taxon category unworked wood worked wood and artefacts charcoal taxon category unworked wood worked wood and artefacts charcoal Alnus sp Prunus spinosa-type Cornus sp Quercus sp. 2-2 Corylus avellana Rhamnus cathartica Euonymus europaeus Rosa sp Fraxinus excelsior Salix sp. Juniperus communis (not Salix repens) Ligustrum vulgare Taxus baccata cf. Ligustrum vulgare / Ulmus sp Lonicera sp Viburnum opulus Lonicera sp Bark remains Pomoideae Indet Prunus sp total = not present Table 3.5 Schipluiden, unworked wood, worked wood and wooden artefacts, and charcoal (Kooistra 2006b; Louwe Kooijmans and Kooistra 2006). The category of worked wood (primarily artefacts) comprises all excavated finds (N = 185 for the period studied, see table 3.5). The worked wood was divided into four groups: implements/tools, waste of woodworking, wattle and posts and stakes. The group of implements contains a bow (Juniperus communis), two paddles (Fraxinus excelsior), fragments of axe hafts (various taxa), straight sticks interpreted as spears or javelins (various taxa) and artefacts with an unknown function. The wood working waste consists of worked branches, wood chips, tangentially split-off rectangles, and tangential and radial planks and split pieces. The wood interpreted as wattle was unworked though it had a small diameter and showed traces of wear and deformation similar to that 91

20 observed in the active elements of wattle work. The investigated posts date to the Middle Neolithic and were all part of the fence that surrounded the site. A third of the post holes of the fence contained wood remains that were preserved well enough for wood identification (N = 91 out of N = 299). The fences were made of six taxa, of which Prunus sp., Juniperus communis and Alnus sp. were used most often. Many posts were made of freshly cut wood. The diameter of the posts was concentrated around cm. Differences in preservation and the presence of post remains between various fences indicate that some fences were left behind (especially posts of Juniperus communis and Alnus sp.) while other fences were probably pulled down, possibly for re-use of the wood. There is no information on the taxa used for houses since it was not possible to reconstruct which groups of postholes formed single houses and since the relevant postholes did not contain wood anymore. Most worked wood (artefacts etc.) dates to phase 2a because of the favourable embedding conditions during this phase. It is therefore not possible to investigate diachronic developments of wood working. It is highly likely that wood working occurred at the site since the taxa used for artefacts correspond with the other attested taxa, since mainly branches and few trunks were used and since sub-optimal taxa were used for artefact manufacture. The remains of phase 2a were concentrated at four locations that probably represent locations of wood working (Louwe Kooijmans and Kooistra 2006). Charcoal was collected by hand during excavation and from 4 mm sieve residues. Data were derived from phase 1-2a, phase 2a, phase 3 and from phase 1-3 (not assigned to a single phase), resulting in 1134 identifications. The analysis included charcoal from the refuse layer at locations evenly distributed over the dune, and charcoal from wells, hearth pits and a single pit. The identified taxa are shown in table 3.5. The charcoal assemblage is dominated by Alnus sp., Pomoideae, Prunus sp., Juniperus communis and Salix sp. It shows high similarity with the assemblage of unworked wood, indicating that the gathering of firewood was in the first place based on availability of taxa in the natural vegetation. There are, however, also some differences with the unworked wood assemblage, since Juniperus communis is relatively scarce in the charcoal assemblage, Rhamnus cathartica is remarkably common in the material of phase 1-2a, and since individual hearths were dominated by Pomoideae, Prunus sp. or Alnus sp. However, the variety of the taxa that are more commonly found than expected and their common presence in the natural vegetation does not support selective use based on the qualities of the wood, and the combustion qualities of individual taxa cannot be related to charcoal selection for burning qualities either, except for Pomoideae. Most of the charcoal remains consisted of branches or twigs, supporting that the fuel was gathered in the exploitation area (Kooistra 2006b, 367). The publication does not contain information on the contents of single wells, or information on the use of moist wood or brush wood affected by fungi. The analysis of wood and charcoal indicates that the wood supply did not change through time. Most of the wood was probably collected in the exploitation area of the site, e.g. in the surroundings of the site, since the taxa used for wooden artefacts corresponds with the assemblages of pollen, macroremains and unworked wood. Selective use of wood for artefacts based on the quality of the wood probably occurred on a small scale only. A positive indication of selective use of wood at Schipluiden consists of the use of Fraxinus excelsior for two paddles. Negative indications of selective use of wood at Schipluiden consist of the use of Juniperus communis instead of Taxus baccata for a bow, the use of branches of various taxa for hafts instead of trunks of only few taxa, and the use of various taxa for the fences. Analysis of the branch thickness and the number of annual rings indicated that there are no indications of re-use of the same groves or initial forms of management of vegetation. The number of annual rings could however not regularly be counted (Kooistra 2006b, 364) and management of the vegetation can therefore not be excluded (Kooistra 2006b; Louwe Kooijmans and Kooistra 2006). 92

21 3.6.7 Other sources The molluscs found at Schipluiden are Mytilus edulis, Scrobicularia plana, Cerastoderma glaucum and Cerastoderma edule. These species indicate calm saline or brackish conditions, although they do not necessarily represent the local environment since they may have been transported from elsewhere naturally during spring tides or by people (Kuijper 2006). The analysis of pottery indicated that the occupants of the site used molluscs as pottery temper (Cerastoderma edule/glaucum and Hydrobia ulvae/ventrosa) but there are no indications of the consumption of molluscs. The insect analysis indicates a very open landscape and does not provide indications of presence of woodland vegetation, dense shrub vegetation, or decaying wood. The number of investigated samples is however limited. The analysis supports that the environment was occasionally flooded during phase 1/2a. The insect remains further indicate the presence of Brassica sp., Capsella bursa-pastoris and Beta vulgaris (Hakbijl 2006). The background fauna indicate the presence of woodland vegetation and wetland conditions in the exploitation area (Zeiler 2006c) Discussion The site Schipluiden was located in the middle of various ecozones: the dunes behind the coast, the estuary and river area and the freshwater marshes. The plant and animal remains demonstrate that people exploited all these ecozones. The sea and the shoreline were however not intensively exploited. The pollen, macroremains and insect assemblages indicate that the dune was very open. It was not covered with deciduous woodland vegetation or dense shrub vegetation, although the presence of small patches of shrubs remains a possibility. The pollen and macroremains data allow the presence of some shrub vegetation during the very early phases, and show an increase in open vegetation through time. The wood and charcoal data are dominated by Alnus sp., Pomoideae and Prunus sp., while Juniperus communis and Salix sp. are additional important taxa. This varied composition and the contrast to the pollen and macroremains assemblages shows that the wood represents the vegetation in the exploitation area instead of the vegetation on the dune only. The dune shrub vegetation (Rhamno-Prunetea) probably grew on calcareous sand soils in the dune environment, generally west of the site, while the alder carr was present on eutrophic and mesotrophic peat soils in the freshwater marsh environment. These peat soils were presumably located east of the site in the parts of the beach plain that were hardly influenced by brackish influxes anymore (from phase 2a onwards). Kooistra (2006b) convincingly argues that Juniperus communis was probably part of this dune shrub vegetation, at least at those patches that were open and where sand was not blown over the vegetation. The fences show that small trees were present in the exploitation area at least from phase 2a onwards. The growing conditions for trees and shrubs that were present in the exploitation area were suboptimal, as indicated by the fact that many artefacts are made of branches and thin trunks instead of thick trunks and by the distance between annual rings of investigated wood remains. The diameters of many postholes however indicate that thick trunks were used at the site as well. Only alder may have grown under more favourable conditions. The waterlogged wood indicates that some deciduous trees were absent or had a limited presence in the region, including Acer sp., Tilia sp., Fraxinus excelsior, Quercus sp., Ulmus sp., Corylus avellana and Taxus baccata (but see the discussion below on the vegetation in the region). Reconstruction of human impact on the vegetation is hardly possible since the pollen analysis does not allow reconstruction of the vegetation at the dune before, during and after occupation due to the character of the natural vegetation and due to the low sample density (Bakels 2006). The indications of the decreased presence of shrubs on the dune can be related to human impact. The herbs furthermore indicate the presence of disturbed, eutrophic and possibly enriched patches on the dune. Human impact probably influenced the vegetation in the exploitation area as well, as suggested by the calculation of wood needed for the fence (Hamburg and Louwe Kooijmans 2006). 93

22 The environment around the site of Schipluiden was suitable for crop cultivation, especially during phase 2 when marine influence was decreasing while the ground water level was not yet too high. It is concluded that crops were cultivated locally, in the surroundings of the site (Kubiak-Martens 2006a). Local crop cultivation at the dune or in the exploitation area of the dune is supported by the presence of chaff remains of naked barley, the presence of potential arable weeds, finds of querns, finds of sickles with sickle gloss in a longitudinal direction resulting from the cutting of cereals, and the site function (a year-round occupied settlement) (Kubiak-Martens 2006a; Louwe Kooijmans 2006b). The shift from naked barley to emmer dominating the crop assemblage also supports the practice of local cultivation in the exploitation area (see paragraphs and ). The arable weeds indicate that the fields that were used by the occupants of Schipluiden were probably not located on the high salt marshes (contra Kubiak-Martens 2006a, 333 and Louwe Kooijmans 2006b, 503) but on sheltered parts of the beach plain or on (the lower slopes of ) dunes. Marine influence may have affected the fields though presumably decreased after the first phase. Further, the weeds indicate that the plots were not shifted frequently. The location of fields on the dune of Schipluiden itself seems unlikely since people needed their space for domestic activities, since chaff and straw is relatively scarce and since there is no evidence of tillage marks. However, scarcity of chaff may be related to use and taphonomy, and absence of tillage features can be explained by site-formation processes. 3.7 Wateringen Archaeology The site of Wateringen 4 (coordinates / ) was excavated in 1994 by the Institute of Prehistory Leiden (now the Faculty of Archaeology). The research methods and results are published in Raemaekers (1997) and Raemaekers et al. (1997). The site was the first Middle Neolithic site discovered in the region. The size of the excavated area is c m 2, and the distribution of finds and features suggest a similar size for the site, although complete excavation of the dune was not possible. The top of the dune is at 3.60 m -NAP. Six stratigraphical units were recognised (see table 3.6). Unit 3, consisting of peat, represents the refuse layer. Occupation of the site took place sometime between 3625 and 3400 BC. unit sediment site formation period 6 clay disturbance by marine sedimentation 5 peat peat formation 4 clay erosion and clay sedimentation from 3350 BC onwards 3 peat with sand lamina peat formation and occupation BC 2 fine sand dune formation and soil formation circa BC 1 calcareous sand and clay beach barrier formation circa 4000 BC Table 3.6 Wateringen 4, stratigraphical units and related processes (Raemaekers et al. 1997). 94

23 Features at the site comprise postholes, some with the remains of posts, unlined wells, shallow pits interpreted as watering places for animals, pits with an unknown function, and a single hearth outside the house. In the postholes a two-aisled house (11 x 4 metres) has been recognised. Finds comprise flint, stone and organic material. The occupation period and presence of a single cluster of postholes suggest that Wateringen 4 represents a single house site used by several generations, estimated at c. 50 years. The pottery is characteristic of the Hazendonk group. It was mainly spread on the top of the dune and around the house, similar to the distribution of the flint. Most of the flint (c. 90%) could have been obtained locally (from southerwestern coastal sources), but the sources indicate contact with the south as well. The stone assemblage comprised querns, but sickle inserts were absent. Use-wear analysis of the flint demonstrated working of siliceous plants (reed or grasses), wood, bone, hide and unknown mineral material. The animal bone assemblage (N = 657) was dominated by cattle, pig (wild/domestic) and red deer. Bones of sheep/goat were absent. Most of the birds and fish indicate a freshwater environment, while a minority is indicative of brackish conditions. The mammal remains indicate that the focus of exploitation was in the coastal plain rather than the freshwater marshes or the sea. The faunal remains (birds and fish included) indicate occupation during summer and in early autumn and winter (between October and March). The investigators argue in favour of year-round occupation because of the suitability of the region for year-round occupation, the presence of a house and the indications of a relative broad spectrum of subsistence activities (Raemaekers et al. 1997) Archaeobotanical materials and methods The archaeobotanical research included the analysis of pollen, macroremains, wood and charcoal (Bakels in Raemaekers et al. 1997, Hänninen and Vermeeren 1995; Hänninen and Vermeeren in Raemaekers et al. 1997). The pollen analysis consisted of two samples with a volume of 1 cm 3 collected from the base (4.02 m -NAP) and the top (3.95 m -NAP) of the refuse layer at 50 metres distance from the top of the dune. The analysis is based on a total pollen sum. The density of pollen was low. The macroremains analysis consisted of 59 samples that were collected by interval sampling (standard samples collected at five metres distance from each other) and 15 samples from features (unlined wells, a watering place, a posthole and pits). The sediment of most samples consisted of (humic) sand. The sample volume varied between 0.5 and 3 litres. The mesh width of sieves was 0.5 mm. The results presented in Raemaekers et al. (1997) represent a selection of the macroremains below (see also the discussion there). A small number of identifications has been revised. 7 The macroremains assemblage is strongly influenced by selective corrosion since mainly resistant macroremains were preserved. The wood assemblage can be divided into unworked wood remains (N = 64), pointed wood remains (N = 15) and worked wood remains (N = 12). Only the assemblage of worked wood remains was investigated completely, while a selection of remains was investigated from the other categories. A part of the unworked and worked wood remains was collected from unlined wells and watering places. The pointed wood remains were partly collected at the location of the house. The charcoal (N = 65) was mainly derived from four concentrations collected from macroremains samples, and a hearth that contained 106 grams charcoal of which 8.8 grams were identified (25 identifications). 7 The identification of Potentilla reptans has been replaced by Ranunculus sceleratus. The identification of Bromus sp. has been rejected. The identification of a carbonised seed of Brassica nigra from sample N94 has been rejected. The identification of Brassica nigra from sample C88 is replaced by Brassica rapa. The identification of Berula erecta is replaced by cf. Apium graveolens. The identifications of Polygonum minor (C88, UU119) are replaced by Persicaria maculosa/minor. Sambucus nigra is replaced by Sambucus cf. nigra. 95

24 3.7.3 Pollen analysis The results of the pollen analysis from the refuse layer are shown in table 3.7. The pollen sum of the sample from the base of the peat is very low and restricts the representativity. Both samples do not indicate the presence of dryland woodland/shrub vegetation on top of the dune, but the presence of shrub vegetation cannot be excluded since pollen of shrubs are generally underrepresented. At the start of the formation of the peat, Asteraceae liguliflorae, possibly representing Eupatorium cannabinum, were present in the local vegetation. The sample from the top of the peat shows increased values of Chenopodiaceae, Poaceae and Foraminiferae, which probably indicate increased marine influence related to the presence of clay above the peat. It is not possible to recognise human impact in the pollen spectra due to the small numbers of samples, the scarce information on the (natural) vegetation and the distance between the sample location and the top of the dune where the house was located Macroremains analysis The results of the macroremains analysis are shown in table 3.8 (at the end of the chapter). The macroremains analysis gives no evidence of the presence of dryland trees in the extra-local vegetation. It is argued that the common presence of Prunus spinosa and Sambucus nigra indicates that these shrubs were part of the local vegetation on top of the dune, although import cannot be excluded (Raemaekers et al. 1997, 156). Presence of shrub vegetation is supported by the presence of herbs indicative of shade and woodland edges, although these base peat top peat base peat top peat Upland taxa Wetland taxa (cont.) Pinus Sparganium erectum-type Tilia Urtica dioica-type - + Quercus Corylus Ecologically indeterminate Betula - + Apiaceae + + Artemisia - + Asteraceae tubuliflorae Chenopodiaceae Cyperaceae Polygonum persicaria-type + - Poaceae Polypodium Poaceae > 37 µm - + Monoletae, psilatae Wetland taxa Triletae, psilatae Alnus Foraminiferae Asteraceae liguliflorae Lycopodium (marker) Calystegia sepium-type - + Euphorbia palustris - + Filipendula Pollen sum = present - = not present Table 3.7 Wateringen 4, pollen identifications of two samples based on a total pollen sum, analyst: W.J. Kuijper,

25 could also have grown in the shade of the house. The more open parts of the dune were probably overgrown with forb vegetation and taxa that are indicative of disturbance of the soil (e.g. Chenopodium album and Persicaria lapathifolia). The distribution of these taxa at the site, corresponding with the distribution of marshy taxa and not corresponding with the distribution of cereal remains, supports their presence in the local vegetation. However, some of the taxa indicative of disturbance may also have functioned as arable weeds (discussed below). Taxa of riparian vegetation, marsh vegetation and moist grassland rich in nitrogen are found in the lower parts around the dune (e.g. Eupatorium cannabinum and Euphorbia palustris). There are no indications of a nearby presence of alder carr or open water. The data from primary and secondary pit fills suggest that the vegetation hardly changed through time (Raemaekers et al. 1997, 151). Taxa that tolerate and prefer moderate brackish or saline conditions are Alopecurus geniculatus, Apium graveolens, Bolboschoenus maritimus, Chenopodium rubrum, Phragmites australis, Schoenoplectus tabernaemontani and Suaeda maritima. The presence of these taxa indicates that minor brackish influxes may have occurred, although the number of taxa that are indicative of true brackish conditions is low. The fruits of Suaeda maritima were concentrated around the house and it is presumed that they were imported with clay used at the site, or that they arrived at the site after consumption by cattle (cf. Raemaekers et al. 1997). The group of taxa found in a carbonised state (see table 3.8 at the end of this chapter) consists of many taxa that together represent various ecological groups: crop plants, potential arable weeds, potential gathered food plants and taxa that represent the natural vegetation including marsh taxa. Potentially gathered food plants found in a carbonised state are cf. Malus sylvestris, Corylus avellana, Prunus spinosa and Rosa sp. Other potential food plants that were found in a waterlogged state only are Cornus sanguinea, Sambucus cf. nigra and Rubus caesius. The assemblage of crop plants found at the site included carbonised grains and chaff remains of Triticum dicoccon and Hordeum vulgare var. nudum. The ratio of both cereals is approximately equal when considering grains and chaff remains. Concentrations of carbonised cereal remains or chaff remains were absent. The potential arable weeds are presented in the synthesis of the region (see paragraph ). The macroremains assemblage from Wateringen 4 contains many taxa that are known as arable weeds. It is however unclear which taxa represent true arable weeds. It is not possible to distinguish weeds by the analysis of concentrations of crop products since the samples that were rich in carbonised remains of cereals (the samples in pit C88) also contain carbonised remains of food plants, potential arable weeds, and marsh taxa or water plants, which suggest that the samples are of mixed origin and do not represent pure products of crop processing. Frequency analysis of taxa found in a carbonised state in samples with carbonised remains of cereals indicates that Prunus spinosa, Galium aparine and Persicaria lapathifolia are most frequently found together with cereals, suggesting that these taxa functioned as food plants or as arable weeds Wood and charcoal analysis The large variety of taxa in the unworked wood assemblage suggests the presence of open dune vegetation, dune woodland vegetation and moist beach plains in the exploitation area of the site. Alnus sp. and Juniperus communis dominated both the unworked wood assemblage and the worked wood assemblage, indicating that these taxa were most easily available in the exploitation area. The outer posts of the house were made of Juniperus communis (diameter 6-16 cm; see fig. 3.4 and 3.5) while the inner posts of the house were made of Alnus sp. (diameter 8-13 cm). The trunks of Juniperus communis indicate good growing conditions. The selective choice of the two taxa for the two types of posts is presumed to be related to the function of the posts and the characteristics of the wood although selection due to a religious and/or symbolic meaning especially of the evergreen Juniperus communis has been suggested as well (Hänninen and Vermeeren 1995). The post of Quercus sp. probably was not part of the house since it was located outside the structure and since the wood taxon does not correspond with the results of the remaining posts (ibid.). There is no explicit information on the 97

26 Figure 3.4 Wateringen 4, post of the house (Juniperus communis), scale 1:2 (photograph: Biax Consult). Figure 3.5 Wateringen 4, post of the house (Juniperus communis), scale 1:2 (photograph: Biax Consult). 98